It depends on the problem. When you are interested in residual stress and deformation of steel showing phase transformation from austenite to martensite, a commercial software Sysweld could be a good resolution to your question.
If you wish to use standard FEA software packages (commercial or open source), the approaches to the analysis of heat flow, residual stress/ distortion are elaborated in the following two short notes (and other associated material) at my page on www.researchgate.net
1. Note on FE analysis of heat flow using point, line, surface and volumetric moving heat source.
2. Note on methods of calculating residual stresses and distortion in welding
I confer with the previous researchers - it depends on your specific welding process, what you want to model (melt pool, residual stresses ...) and the scale you want to model it at. For the melt-pool, FLUENT can be used, but it is not straightforward. For residual stress you can also use ANSYS-APDL or workbench Thermo-mechanical using element birth and death.
It is true that the method of analysis depends very much on the objective of study. If micro-structure changes are the objective of study and the material is susceptible to what happens in the molten state, melt pool analysis is necessary. On the other hand, if the objective of study is to evaluate residual stresses (for establishing stress relief strategy) or to evaluate distortion (for controlling dimensional changes), the melt pool can be disregarded and broader temperature distribution with subsequent elasto-plastic analysis are more important. This too becomes unmanageable when the structure is complex and to-date only approximate mixed methods (an intutive simple analysis in weld bead+FE analysis in rest of structure) have been tried with different degrees of success.
Since the question was so open ended I would like to put some bounds on it. I run FEA with AutoDesk Simulation (Algor) on a regular basis. The FEA is of large tanks that contain a fluid and generally has large flat plates welded together, with stiffening welded to the large plate. These tanks experience a pressure cycle. I've looked at three methods to model the weld goemoetry: 1. Round line elements of with a diameter equal to the mesh size, 2. plate elements with a thickness equal to .707 x weld leg length, and 3. solid elements of the ideal weld geometry. Each has an increasing complexity and adds processing time. In your opinion, which method would provide the least error with the least processing time? Or, is there a method not listed that would work better? The results should accuratly estimate the stress in the plates and therefore can be used to asses the risk of a leak due to low cycle fatigue.
Algor is a general purpose FEA software which can be used for routine design and stress analysis. But when it comes to the stress analysis for special purpose like the crack growth due to fatigue loading in welded structure, I am afraid that a very special approach (semi-analytical with empirical input) is required. The phenomena is so much dependent on the uncertain factors like residual stresses, weld imperfections, internal/ external weld defects, material characterization etc. that a numerical analysis in isolation of the input of practical information is, i am afraid, futile. I have some information about the fatigue/ low cycle fatigue behaviour of welded joints, which I am sharing with you.
First is a document produced by Prof. A. Hobbacher for Int. Inst. of Welding, which is a compilation and final recommendation from research of many contributory researchers. It is
'Recommendations for assessment of weld imperfections in respect to fatigue' by A. Hobbacher , IIW doc. XIII-1266-88/ XV-659-88
Second is a reference to the British Standard on low cycle fatigue of welded joints and its recommendations in the book Welding Engineering and Technology by R.S.Parmar
I have not tried directly the FE analysis of this problem but the following explanation may be useful. Induction can be modelled using commercial software as well as some open source software (such as Elmer). The induced current can later be used to determine heat generation (Joules’ heating) and subsequent thermal analysis is used to determine temperature. All these types of analyses can be coupled in FEA software. In composites the problem is to obtain magnetic, electrical and thermal properties, because it has two components (polymer + fibre – Carbon etc) with widely different properties. It is to be ascertained whether some averaging of properties can be done. If so then the analysis through standard software modelling can be tried. In this case subsequent experimental verification is a must.
As suggested by Muthanna Abbu, Comsol Multiphysics is a good choice to model the welding process. This allows the implementation of different physics using user friendly application models. But this definitely depends on the objective of your study as described by Dr. O. P. Gupta. I have successfully used this for the weld pool flow study during static spot welding process.
Simufact.welding is a dedicated welding FEA software, with very easy to use interface. It uses volume elements and implicit time integration for high accuracy.
You can simulate either thermal only, or thermal-coupled processes.
There are many aspects of the problem when you compare bolted and welded joints. One is material, i.e. whether it is fracture sensitive or ductile. Second, if the micro-structure undergoes change during welding, it will influence the weld strength. Third is the general quality of the weld (imperfections etc.). Several other aspects, such as stress concentration, will also play role in it.
Assuming naively that the weld is very sound and has full penetration, the load carrying area of plate is always more in case of welds compared to bolted plate which has reduced area due to bolt holes
ABAQUS with user subroutine DFLUX can be used. However, it is not straight forward in ABAQUS, experienced user can work. On the other hand, there welding simulation software available like SYSWELD or SIMUFACT.